Quality of Service Adaptation in Data Stream Management Systems: A Control-Based Approach

Unlike snapshot queries in traditional databases, the processing of continuous queries in Data Stream Management Systems (DSMSs) needs to satisfy user-specified QoS requirements. In this paper, we focus on three major QoS parameters in a DSMS environment: processing delay, querying frequency and loss tolerance. To minimize processing delays, the Earliest Deadline First (EDF) CPU scheduling policy is recommended. QoS control for data streams is a difficult problem because both the processing cost and data rate can be unpredictable. One salient problem is that DSMSs are vulnerable to overloading as the data points tend to come in bursts. QoS degrades significantly when the system is overloaded. We propose a QoS adaptation framework that smartly adjusts application QoS and performs admission control based on the current and historical system status. The design of the QoS adaptation framework takes advantages of techniques from the area of control theory. Algorithms derived from control theory are mathematically analyzable and this makes the parameter tuning an easy job. In addition to load shedding, our QoS adaptor adjusts system load by changing multiple QoS parameters for streams. Extensive simulations using both synthetic and real data sets are Permission to copy without fee all or part of this material is granted provided that the copies are not made or distributed for direct commercial advantage, the VLDB copyright notice and the title of the publication and its date appear, and notice is given that copying is by permission of the Very Large Data Base Endowment. To copy otherwise, or to republish, requires a fee and/or special permission from the Endowment. Proceedings of the 30th VLDB Conference, Toronto, Canada, 2004 carried out to evaluate the performance of the proposed control-based QoS adaptation framework. The results show that: compared to current solutions for QoS control in DSMS, the control-based strategy provides better guarantees of QoS and achieves higher resource utilization. Our solution is also light-weight and easy to implement.